On Jun 17, 2010, at 8:41 PM, Charles Srstka wrote: > On Jun 17, 2010, at 8:19 PM, Matt Neuburg wrote: > >> On or about 6/17/10 5:56 PM, thus spake "Roland King" <r...@rols.org>: >> >>> if ... you just have a class variable called <key> or _<key> then the >>> whole array is not replaced, the method you call on >>> NSMutableArrayValueForKey: >>> is passed through to that underlying variable which does what you might >>> expect >>> and operates on the array variable directly. >>> >>> That's not such a rare case. >> >> In fact it is exactly the case I had in mind. :) That's really just what I >> wanted to hear. Someone asked "how do you think the proxy object works"; >> that *is* how I was thinking it might work. Thx - m. > > The problem with things like this is the way it destroys encapsulation.
Exactly. Matt, if you have a to-many relationship property, do you make the return type of the getter NSMutableArray*? I should hope not, or you're asking for problems. Therefore, why did you "want[...] to hear" that KVC would do the rough equivalent? It's a terrible idea to let KVC access instance variables directly. My firm belief is that Apple only maintains that "feature" for backwards compatibility. I never subclass NSObject directly. Instead, I have a custom subclass of NSObject one of whose main purposes is to override +accessInstanceVariablesDirectly to return NO. Likewise, whenever I subclass other framework classes, I install such an override, too. Because of that, I mentally excise the direct-ivar-access steps from the KVC accessor search patterns. Regards, Ken _______________________________________________ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
